ivan wrote:Desktop is being worked on, but not in the sense it was around 4.0. Most of the time it fits its purpose, so there's no point in reinventing the wheel again. There *are* some things that need improvement though, but I digress.

The idea of popup SAL was something I've thought of as soon as Marco implemented it, and it would (IMO) be a quite good replacement for Kickoff.

But, and I can not stress it enough, *not* fullscreen. The fact that osx did it, and later others as well doesn't mean it is a good choice. It is ok for smaller screens, not for big ones.

The thing that has always terrified me is the desktop of an ordinary windows user. The whole screen covered in icons, where you can not find anything. And that was in the era of 14" screens.

See: http://www.enterprise-dashboard.com/img/ad-agency-ceo-desktop.jpg for an example.

Now imagine a 23" screen filled with icons.

See http://linuxnow.ru/files/81/unity3.png and imagine how much you need to move your mouse in order to run FTP program.

As soon as you have the need to keep your favourite applications in the panel (like all the mentioned systems do), it means that the application launcher has failed to provide a quick way of launching apps.

Unity not works only in full screen mode but can be resizable to fill only a percentage of the display.

I also use osx and I have a 23" monitor, I haven't problems to move the mouse to click on the application icon, the bad of osx's solution is not to arrange the icons applications by category. Instead in S&L are present and can be filtered by category, this is very usefully and confortable for user.

There must be a reason why the new UI-Menu of the main DE\OS are similar (Unity, Gnome Shell, Metro, Launchpad....)...and the most ironic of all is that KDE had done and realize this first of all with S&L.

First of all, as I said SAL is rather good, so I have nothing against it apart from it being fullscreen.
- provides favourite apps
- browsing
- and searching

> Unity not works only in full screen mode but can be resizable to fill only a percentage of the display.

I know, forgot to mention it as a positive example.

> I also use osx and I have a 23" monitor, I haven't problems to move the mouse to click on the application icon,

Well, it is not a /problem/ to move the mouse, the problem is that you need to. I'm to lazy to draw at the moment, but statistically you keep your mouse around the middle of the screen most of the time. In order to launch something (without using keyboard shortcuts) you first need to drag your mouse to the upper-left corner (approx 11" ~ 2x" / 2), then move it to the app you want to start (approx 11"). So, in essence, the average mouse movement for these kinds of menus equals to the screen size. Which is too much.

The problem is also that you have a 2D grid of *a lot of* items that you need to visually parse while searching for a desired application. It works if the icons have enough visual distinction between them, but it stops working as soon as you need to read the text below the icon.

> There must be a reason why the new UI-Menu of the main DE\OS

Apart from the obvious reason of UIs copying each other, these are the benefits:
- a lot of items an fit a small screen (every mobile phone I can think of in the past few years used this approach for this reason - even before touch screens) - so, good for mobile phones, netbooks, tablets
- great for larger-than-phone touch-based devices - for touch devices, you don't have a mouse cursor, you have either two hands (you hold the device usually with two hands), or it is a big screen in front of you and every point on the screen has the same accessibility - this removes the travel distance issue

Just see where are the UIs in question coming from - iPhone -> OSX, WP7 -> Windows.

A better UI Experience is not the time to move the mouse to find an application but the semplicity to find (visually) this.

With classic menu like kickoff you must:
a)click on icon menu;
b)click on category;
c)scroll the category to find the icon;
d)click on icon to start the application.

There is 3 click and 1 possible scroll.

I also use S&L in 23" and in my 32" TV, I have not problems.

I'm not defending Kickoff, totally wrong person to do so

Maybe most important is to keep in mind this:
For more success of KDE/Linux it could be advisable to keep KDE configurable so users can choose between different ways...
Some like the Windows look and feel. Others (like me), for example, are allergic against iconitis and prefer different ways to do things and how to populate the screen.
Nobody likes to be forced to do it another way one likes it.

There should possibly be a few preconfigurations not only in the looks (colors, themes etc) but in the feeling, too.
For example, Imagine a configuration/installation question:
What system do you want?
1. Windowish
2. Macish
3. Unixish

which configures many things not only in the looks (icon view vs. detail view, etc) but in the feeling too (click to focus to focus follows mouse etc) and all these things inbetween (windows-95-ish start menu bar to twm-ish popup launcher).
Maybe people could feel instantly "home" without having to spend hours of detail configurations?

Feline: kinda like Chakra does it? When you install it you get a choice between different GUI set-ups which is pretty slick actually. Another thing would be a simplified Visual Settings Manager (where you chose like a Mega Theme first and then go in and edit the details).

Ivan: personally I am all for the full screen thing. As long as its obvious to any user that this is a temporary space. That this is something else than the normal desktop and that you have a given, simple way out of it. Gnome solves this by darkening the desktop, and always having the "application button" present, the major details of the desktop are the same BUT it works in a different way. The issue with Windows 8 was that they simply lacked a simple method of moving between instances and methods of launching. The user was stuck in a different "desktop landscape" without any real knowledge of how he/she got there.

Its like when you switch a desktop - my father-in-law did it by mistake in LXDE and couldn't for the life of him understand where all the things he'd been working on went. But when I showed him what happened on my laptop (KDE) with a sliding animation - he completely grasped what had happened and could look for a sollution. The simple little animation told him what had happened, where he "where" on the desktop landscape (its a word now, I'm making it an actual word).

One sollution a median between the full-screen-launch and the pop-up is broadening the pop-up making it essentially "full screen" BUT still not. Instead of making it full screen visually - you make it "full screen" in actuality (much like the Kickoff launcher which is essentially "full screen" - if you click anywhere beyond it, it disappears. You can only interact with it and it alone). To show what has just happened you make it half-screen or 3/4 visually, darkening the desktop and pushing it upwards (or whatever direction its coming from). Just like the Takeoff launcher when you start typing it starts a search.
BUT unlike the takeoff you can narrow the search depending on where you are and defined by little "area" icons (Home/Favorites, Applications, Files and Folders and System). The main difference from, for example, the kickoff launcher is that visually it simplifies the launch procedure by removing options from scratch that are unused or can be accessed elsewhere just as easy AND it shows the user the unique position on the Desktop Landscape they are in. Unlike a full screen launcher it doesn't throw the user into a new environment with unknown rules for how the system works but shows what have just happened. "Where they are" as it where.

I made some very very hastily done examples to show what I ment. Ignore design details and stuff like that, this is just an example.

Before launching:

After launching and when you start typing:


As a side note: I think we should always strive to "reinvent the wheel". Otherwise we would be standing around trying to roll stuff on large logs instead of vulcanized tires. Nothing is ever complete.

The issue which Ivan mentions - the long mouse travels one has to do with "start menu buttons" is really annoying.
(Call it Kick-whatever if you like - its' actually no more than a clone of Redmond "invention"...)

The current KDE start menu implementation is even more annoying than the Redmond one: you have to click the right arrows to get into the deeper menu tree levels. These should pop-up tree-like when you hover around the menu, as it was in early Windows95. The current KDE way is an awkward implementation just to distinguish itself from the original.

I also prefer Ivan's suggestion: pop up the start menu when you click on a free desktop spot. This is more straightforward context-sensitive and relieves one to be forced to always travel to a particular screen location.

The second image Jens posted I find a great idea:
in the left of the popup are several menu tree roots (applications, file, system etc) and on the right there are the icons of the programs the user _actually_ uses.

Windows 7 for example memorizes the few last recently used applications and shows them when the start menu is opened.
However, the number of "menu slots" for the last recently/mostly used programs is very small.
It may be sufficient for office workers with only two or three oft-used programs.
However for power users this is insufficient.
If one regularly uses a set of, say, fifteen programs, this feature in the Redmond start menu is rather useless, as it never can show all programs used.

So, if the list of icons marked "icon" in Jens' example is being populated by the programs the user _actually_ (!) uses, they would be a real shortcut.
In the beginning this list would be completely empty, no "preselected" programs, and over the time filled by the most-used programs.

This automatic approach would save one from the tedious task to edit the start menu for having the most-used programs in the first level.
And there would be no need to skip through macish-androidish-gnomish full screens of of never-used programs' icon garbage anymore when visually parsing the menu.

Together with a kind of "mega-meta setting" which Jens mentioned, allowing users to easily set the general GUI behavior direction without hours of configuring a plethora of details, this would be like a newly invented, improved wheel, after the invention of tyres.
If KDE would make this leap it even could overcome the need for a start menu bar, to get a distinct, unique and comfortable look and feel, instead of just mimicking Microsoft Windows. It would stop making the impression of just being a cheap clone.
This would be my vision how next generation KDE works!

Modernity: I am in the camp that "Modern" full screens don't make sense for desktops. OS X's Launchpad is not the primary or generally accepted way of accessing applications. Having a full screen icon grid is just a waste of space. It is an iOS-like port in addition to the normal application folder. Windows 8's Start Screen has been such a huge failure that people are celebrating at rumors that the Start menu is coming back. It should go without saying that taking "Modern" cues from Gnome 3 or Unity should be given a skeptical eye, given the community's reactions to them. Given the influence of modern OS on our aesthetic, I think icon focus is fair choice. But full-screen is a leap.

Proposals thus far: While I think there is room for improvement, I don't think the "half screen + slide away" implementation quite solves the problem of removal of context. The top half of the screen is now essentially non-space, so this in fact adds a "problem" such that the Gnome implementation actually seems superior here. And you have to wait to "exit" the menu and the desktop to reemerge to click on a window and resume your workflow.

My suggestions: I just don't see the utility gained by isolation, which is the number one source of complaint I have heard about these launchers. Why is making big icons with big, empty spaces a goal? Content should dictate form, IMO. On OS X the Applications folder is an icon grid (by default), but of reasonable proportions. Though obviously much too basic. You could add tabs at the bottom, the first and default open one being "Applications". And then applications could be categorized, perhaps like in the second picture (though obviously more reasonably packed)? The bottom row being favourites. This could then rotate depending on if the Plasma widget is anchored in a panel on the bottom, top, left, or right.





The default KDE sub-categorisations are a bit too much, IMO, especially with the default launcher thatt requires so many clicks to navigate. The default should be fewer categories, with users adding more on their own when they feel they need to, instead of all new users being encumbered.

I agree that the user experience would be greatly improved by reducing the number of categories and carefully considering where things belong. Being relatively new to KDE, I've spent more time than I care to admit 'exploring' the different categories. A full install of the kde4 suite was bewildering. I'm not sure that's really about the launcher itself, though.

"Reinventing the wheel" is only useful if the new invention ends up superior. Too many times people reinvent the wheel to everyone's annoyance.

Personally I don't use kickoff because I can't stand the animations. They get in the way enough that it drives me nuts. I use Unity at work, and find the unity dash to be wonderfully quick to use (after disabling all the web search crud). To mimic that I just use krunner with a shortcut. That said, I think there's value in having something similar to kickoff be the default because it helps new users. Once you know what you're doing, you can remove it.